October 16, 2012
Querying Intensional Data
Pierre Senellart
2 / 3 Eilat Workshop Pierre Senellart
Intensional data is everywhere
Lots of data sources can be seen as intensional: accessing all the data in the source (in extension) is impossibleor very costly, but it is possible to access the data throughviews, with someaccess constraints, associated with someaccess cost.
Indexesover regular data sources
Deep Web sources: Web forms, Web services
The Web or social networks as partial graphs that can be expanded bycrawling
Outcome ofcomplex automated processes: information extraction, natural language analysis, machine learning, ontology matching Crowd data: (very) partial views of the world
etc.
3 / 3 Eilat Workshop Pierre Senellart
What we need for querying intensional data
Answering queries using views
Query feasibility underaccess restrictions
Recursive languages: iterating over answers, feeding back an output value as input to a service, repeated accesses to similar sources Plans and cost models for good sequences of accesses
Uncertainty management: the views may be uncertain; the outcome of an access can be modeled as an uncertain process
Provenance for correlation management and explanation Gatheringstatistics about the behavior of the sources